This is what AI is not good at

I ran a little experiment a few months back that showed me exactly where AI falls flat on its face.

And I say this as someone who uses AI every single day, who's built half his marketing ops on it, and who'd rebuild it all again tomorrow if I was starting again (if you wanna see the kind of things I'm building, check this one out...)

At the time, I was trying to grow the email list for a publishing brand I was running called Vagrant Press. Everyone kept telling me AI is basically a strategist in your pocket now.

So I tested it. Properly.

I opened the latest model of Claude, gave it the full picture of the business, audience, and offer.

I then outlined my resources and asked it to do deep research and devise a plan on the best ways to get the next 1,000 email subs.

Fair test, right?

A clear goal, a real business, all the context it could want, and me genuinely happy to do whatever it said.

Here's what the "strategist in my pocket" came back with...

... A free lead magnet, modelled on a Digital Marketer campaign that pulled 350,000 leads. Sounds incredible. Til you check the case study and see it's from 2013, back when Facebook ads were cheap as chips and AI hadn't made 99% of free lead magnets obsolete.
... A giveaway. Run 'em before. The leads were the lowest-value leads ever.
... A referral programme. For a list that barely existed. Who's referring who, exactly?

Think about the last time you asked ChatGPT or Claude for ideas or strategy. I'd bet good money you got a similar surface-level response based on ideas that worked 5+ years back, but don't anymore.

There's a reason for that, and it's built into how these models work.

They're just pattern recognition machines, and they prioritise the things with the most mentions. So, the ideas and approaches that have been covered the most, get surfaced as the "best" options.

And in marketing, the advice that's been written about the most is the advice that's already been done to death.

By the time a tactic shows up in enough blog posts and YouTube videos for AI to learn it, the edge is gone.

So when you ask AI for strategy, it defaults to the most popular stuff other people were running 5+ years back.

Which obviously ain't gonna work.

And this, right here, is why so many businesses are using AI and staying busy without being productive.

They're executing yesterday's playbook. And sure, they're doing it quickly, but they're doing it in a time when that playbook is next to useless.

Look, what you have to understand about AI is that It's nothing more than a catalyst.

Feed it shit, and you get shit faster. Feed it a proper strategy, and the good stuff comes faster instead.

What it can't do is come up with that strategy.

The WHAT to produce and the WHY, the stuff that's become so scarce nowadays I've been banging on about all week.

This stuff has to come from your business, your customers, your numbers... but most importantly your brain. That's the one thing AI is never gonna be able to fully replace.

Let AI do your thinking for you, and you're gonna spend more time trying to get it to a halfway decent idea.

But have a real person driving the strategy, and demote AI to implementing the grunt work, and things tend to go pretty damn well.

The trick is in finding the right strategy and model for your business. Something that's proven, but adapted to your specific needs in a way others in your space are not doing/thinking about.

Get that, and your're on to a winner.

More on this tomorrow.

Pete "2013 called" Boyle